Apple inc. (20240338556). DYNAMIC VARIABLE BIT WIDTH NEURAL PROCESSOR simplified abstract
Contents
DYNAMIC VARIABLE BIT WIDTH NEURAL PROCESSOR
Organization Name
Inventor(s)
Paolo Di Febbo of Sunnyvale CA (US)
Waleed Abdulla of Mountain View CA (US)
Chaminda N. Vidanagamachchi of San Jose CA (US)
Yohan Rajan of Cupertino CA (US)
DYNAMIC VARIABLE BIT WIDTH NEURAL PROCESSOR - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240338556 titled 'DYNAMIC VARIABLE BIT WIDTH NEURAL PROCESSOR
Simplified Explanation: The patent application describes an electronic device with a neural processor that can operate in multiple modes of different bit widths. The device includes neural engine circuits that can perform computations using different kernel coefficients.
- The neural processor has multiple neural engine circuits operating in different modes.
- Each neural engine circuit includes multiply circuits for generating output data.
- The device can switch between modes to perform computations with different bit widths.
- The neural processor can combine multiply circuits to work together in a combined computation circuit.
- Different kernel coefficients are used in each mode for the computations.
Key Features and Innovation:
- Neural processor with multiple modes of operation.
- Neural engine circuits with multiply circuits.
- Combined computation circuit for efficient processing.
- Ability to switch between modes for different computations.
- Use of different kernel coefficients for varied operations.
Potential Applications: This technology could be used in:
- Artificial intelligence systems
- Machine learning applications
- Signal processing devices
- Robotics and automation systems
Problems Solved:
- Efficient computation with different bit widths
- Flexibility in neural processing tasks
- Improved performance in various applications
Benefits:
- Enhanced processing capabilities
- Versatility in handling different tasks
- Optimized performance in neural computations
Commercial Applications: Potential commercial uses could include:
- Neural network accelerators
- Edge computing devices
- Smart sensors and IoT devices
- High-performance computing systems
Questions about Neural Processor Technology: 1. How does the neural processor handle computations with different bit widths? 2. What are the advantages of using multiple modes in neural processing?
2. Another relevant generic question, with a detailed answer: How does the combined computation circuit improve the efficiency of the neural processor? The combined computation circuit allows the neural processor to utilize both multiply circuits in tandem, enhancing the overall processing power and efficiency of the device. By combining the outputs of the multiply circuits, the device can perform complex computations more effectively, leading to improved performance in various applications.
Original Abstract Submitted
embodiments relate to an electronic device that includes a neural processor having multiple neural engine circuits that operate in multiple modes of different bit width. a neural engine circuit may include a first multiply circuit and a second multiply circuit. the first and second multiply circuits may be combined to work as a part of a combined computation circuit. in a first mode, the first multiply circuit generates first output data of a first bit width by multiplying first input data with a first kernel coefficient. the second multiply circuit generates second output data of the first bit width by multiplying second input data with a second kernel coefficient. in a second mode, the combined computation circuit generates third output data of a second bit width by multiplying third input data with a third kernel coefficient.
